I've been tweaking two different amp boards: one based on the TPA3118 and one based on the TDA7492P over the past few months, I would really like to use ceramics on the output filter to save space and stay consistent with my assembly tech. I am using film caps as input filters currently as they are directly in the signal path, but I am wondering how critical they are for the output filter, especially for the TPA3116 which already has ceramic bootstrap caps per recommendation from the datasheet.
C0G caps can be had in the 100nF range, which would enable them to be stacked in order to get the required capacitance for my application, however, at that range they only reach about 25V, this is technically sufficient for the application, which will be using a ~12V ps. However, I could also get X7R capacitors in 250+V range. I'm wondering if that would be better.
C0G caps can be had in the 100nF range, which would enable them to be stacked in order to get the required capacitance for my application, however, at that range they only reach about 25V, this is technically sufficient for the application, which will be using a ~12V ps. However, I could also get X7R capacitors in 250+V range. I'm wondering if that would be better.
X7R capacitors will likely 'sing' as output caps but otherwise work fine provided they have high enough voltage rating. I'd be inclined to go for at least 100V rating, the higher the voltage the lower the ESR which will lead to lower losses and cooler capacitors. 250V rating would most likely be X7T (slightly poorer temperature stability than -R) but pretty much the best choice for their lowest losses.
I should add that all type II and III dielectrics (i.e. not the C0Gs) have a non-zero voltage coefficient meaning they're going to add distortion on a sinewave measurement. Whether this matters in practice is an interesting question.
I should add that all type II and III dielectrics (i.e. not the C0Gs) have a non-zero voltage coefficient meaning they're going to add distortion on a sinewave measurement. Whether this matters in practice is an interesting question.
Last edited:
The bootstrap capacitors are there to perform a completely different job to the capacitors in the output filter. The BS caps are only acting as an energy storage device and are needed to provide the potential difference that allows the MOSFETs in the output stage of the amplifier to switch properly. In other words they aren't directly in the signal path.
The capacitors in the output filter form a relatively high frequency 2nd order lowpass that filters out the high frequency switching component of the class D amplifier. These are in the signal path, although indirectly, as they form a shunt to ground.
As abraxalito above says the X7R type caps have a non-zero voltage coefficient which basically means the value of the capacitor changes depending on what the voltage across them happens to be. If you use X7Rs in a filter that operates at audio frequencies what you typically see is low distortion over the range where the filter isn't doing anything, but as it starts to attenuate, distortion starts to rise.
The output filter of the TPA3118 operates just within the upper end of the audible spectrum and if appropriately tuned to the impedance of the loudspeakers, wont actually attenuate the signal within the audible band at all. By comparison the distortion of the TPA3118 rises quite a lot anyway towards high frequencies anyway, so even if the X7Rs were to add distortion the amplifiers own distortion could quite easily swamp it.
The simplest answer is probably that TI actually use ceramic caps, X7Rs, or similar, in their evaluation modules for the TPA3118. If these were going to harm the overall performance of the amplifier in any significant way then they wouldn't be using them.
Of course they also use ceramic caps (as far as I remember) as coupling caps before the TPA3118 and these are directly in the signal path. But all things considered the amplifiers own distortion would completely the swamp that of the X7Rs, especially if the corner frequency chosen was in the single digit Hz.
The capacitors in the output filter form a relatively high frequency 2nd order lowpass that filters out the high frequency switching component of the class D amplifier. These are in the signal path, although indirectly, as they form a shunt to ground.
As abraxalito above says the X7R type caps have a non-zero voltage coefficient which basically means the value of the capacitor changes depending on what the voltage across them happens to be. If you use X7Rs in a filter that operates at audio frequencies what you typically see is low distortion over the range where the filter isn't doing anything, but as it starts to attenuate, distortion starts to rise.
The output filter of the TPA3118 operates just within the upper end of the audible spectrum and if appropriately tuned to the impedance of the loudspeakers, wont actually attenuate the signal within the audible band at all. By comparison the distortion of the TPA3118 rises quite a lot anyway towards high frequencies anyway, so even if the X7Rs were to add distortion the amplifiers own distortion could quite easily swamp it.
The simplest answer is probably that TI actually use ceramic caps, X7Rs, or similar, in their evaluation modules for the TPA3118. If these were going to harm the overall performance of the amplifier in any significant way then they wouldn't be using them.
Of course they also use ceramic caps (as far as I remember) as coupling caps before the TPA3118 and these are directly in the signal path. But all things considered the amplifiers own distortion would completely the swamp that of the X7Rs, especially if the corner frequency chosen was in the single digit Hz.
Seems like the consensus is that the output capacitors, as long as they are out of the audible range, should not impose any audible distortion. I really appreciate this information, thanks a ton.
I will add that, as my name reflects, I really like the sound of the ta2020. From designing amps and talking to people and reading this forum, I have learned that distortion can be really good depending on the type of distortion, the speaker and the listener. I'll use x7r caps in the output filters of my future amplifier designs and report back to this thread.
I will add that, as my name reflects, I really like the sound of the ta2020. From designing amps and talking to people and reading this forum, I have learned that distortion can be really good depending on the type of distortion, the speaker and the listener. I'll use x7r caps in the output filters of my future amplifier designs and report back to this thread.
Last edited:
- Status
- Not open for further replies.